Author + information
- aHammersmith Hospital, London, United Kingdom
- bInternational Centre for Circulatory Health, Imperial College London, London, United Kingdom
- ↵∗Reprint requests and correspondence:
Dr. Justin E. Davies, Hammersmith Hospital, Imperial College London, Du Cane Road, London, W12 OHS, United Kingdom.
The possibility of noninvasive estimation of functional lesion severity using computed tomography (CT) recently became a holy grail for cardiologists with the proposition of CT-derived fractional flow reserve (FFR) (1). CT-derived FFR claims improvement over pure dichotomous anatomic assessment (50% diameter stenosis) but has unfortunately failed to numerically match invasive FFR in recent studies, despite improvements in technology (2).
In this issue of JACC: Cardiovascular Interventions, Kim et al. (3) propose a novel and simpler alternative noninvasive methodology for determining lesion significance using CT. By incorporating anatomical parameters of lesion severity such as minimal luminal diameter (MLD) together with an estimation of the subtended myocardial mass by the stenosed vessel (fractional myocardial mass [FMM]), they replicate the findings from CT-derived FFR studies by demonstrating superior diagnostic metrics over anatomy alone. Compared with diameter stenosis, an index derived as the ratio of FFM to MLD showed overall superior correlation (0.61 vs. 0.49) and categorical agreement (79% vs. 60%) with invasive FFR.
The proposition of FMM has several merits. First, it is founded in widely established phenomena that govern blood flow in biological systems: more proximal vessels subtend a larger tissue mass with a higher blood flow rate. Although indirectly derived from anatomy, FMM is largely a physiological index that ultimately reflects the amount of transstenotic blood flow. It is not surprising, therefore, that Kim et al. (3) observed that for any given diameter stenosis, lower FFR values are encountered in stenoses with higher FMMs. Second, FMM brings simplicity into a complex field. By avoiding the need for precise 3-dimensional reconstructions of coronary computed tomographic images and complex computational fluid dynamics calculations (4), FMM reduces the biological noise intrinsically associated with any model-based index. Finally, and perhaps most important, derivation of FMM does not depend on the estimation of maximal hyperemia, a largely variable and unpredictable component of CT-derived FFR. The more simplified functional component of FMM has potential clinical implications, as it allows practically any computed tomographic image to be used; impressively, only 1 of 729 stenoses was excluded for technical reasons, a clear practical advantage over models on the basis of computational fluid dynamics, which excluded up to 13% of patients in recent studies (2).
Trumping anatomy, however, is not necessarily a formidable achievement for any new noninvasive modality, despite the perceived advantages of incorporating a functional component into CT. Using CT-derived 50% diameter stenoses as a comparison does not reflect current clinical practice, as neither CT- nor angiography-based decisions are made using such a simplistic approach. The argument by Kim et al. (3) that by offering superiority over a 50% diameter stenosis cutoff, FMM helps us understand the anatomic-functional discrepancies of coronary artery disease is a little less overwhelming. Instead, the main question to be answered is, how does FMM compare with CT-derived FFR, the mainstream functional computed tomographic modality? Although Kim et al. (3) provide no direct comparison between modalities, some insight can be gained from their comprehensively reported data. First, it appears that their study sample is compatible with previous FFR registries and therefore representative of clinical patients undergoing functional assessment in daily practice (5). With a median FFR of 0.83 (interquartile range: 0.74 to 0.90), they demonstrate a typical histogram of physiologically intermediate stenoses. This contrasts with previous CT-derived FFR studies, which showed a median FFR close to 0.88 (interquartile range: 0.79 to 0.94), as a result of the inclusion of more “normal” stenoses (1,2). By correctly choosing to study FMM in a typical clinical population, Kim et al. (3) inadvertently put FMM at a disadvantage, as ischemia tests are less likely to be accurate when stenoses are of intermediate severity. However, to marked credit to Kim et al. (3), a per range analysis of FMM accuracy against invasive FFR was presented, a plot that gives readers a more universal, sample-independent assessment of FMM’s performance. And judging by the presented V-shaped plot, the pattern of accuracy of FMM is very similar to that of CT-derived FFR: it appears to be able to identify correctly stenoses at the extremes of disease severity (normal and diseased) but fails to match FFR in the intermediate zone, close to the cutoff of 0.80. Future studies will have to address this limitation by establishing the range of values within which FMM is accurate enough to exclude or confirm ischemia.
Despite offering potential benefits over models on the basis of computational fluid dynamics, the methodology used by Kim et al. (3) has limitations worth discussing. First, using MLD as the only anatomical parameter of disease severity grossly underestimates the complex mechanisms of energy dissipation through coronary stenoses and ignores the role of morphological characteristics of lesions in generating pressure drop. Second, using FFR as a reference standard for ischemia is an acceptable pragmatic approach that reflects current clinical practice but a strategy limited by its inability to reflect overall perfusion, coronary flow, and microvascular function. As a result, the relationship of FMM/MLD with invasive FFR still has unacceptably large numeric scatter, which is particularly relevant for intermediate FMM/MLD values and means that FMM/MLD and FFR cannot be used interchangeably for clinical decision making. Future studies will have to address such limitations and should perhaps use other perfusion modalities as discriminators.
It is likely that the use of diagnostic computed tomographic coronary angiography will increase dramatically over the next decades, and the cardiology community should welcome any sound attempt to incorporate physiology into its output. It will take us some time to understand how best to interpret such functional measures and also to evaluate what degree of diagnostic accuracy should be clinically acceptable. For the time being, it is clear that intermediate values of CT-based indexes such as FMM/MLD and CT-derived FFR do not match invasive FFR reliably enough, making the role of invasive assessment for the search of vessel specific ischemia in such patients very compelling. The era of certainty of noninvasive virtual percutaneous coronary intervention planning with accurate predictions of post-intervention FFR results is still a far-flung dream.
As with all diagnostic tools used in cardiology, clinicians will be the arbiters of success. The life and death of new technologies depend on many factors, including their accuracy, ease of use, reliability, and probably most important nowadays cost-effectiveness. If these objectives can all be met, it is likely that CT-based noninvasive physiological indexes will have a large role to play in lesion assessment in the next 20 years. Certainly FMM/MLD takes a large step forward toward simplification and potential broader clinical applicability.
↵∗ Editorials published in JACC: Cardiovascular Interventions reflect the views of the authors and do not necessarily represent the views of JACC: Cardiovascular Interventions or the American College of Cardiology.
Dr. Davies is a consultant to and a holder of significant academic grants from Volcano Corporation and Medtronic; and he is the holder of licensed intellectual property. Dr. Petraco has received travel support and is on the Speakers Bureau of Volcano Corporation.
- American College of Cardiology Foundation
- Koo B.-K.,
- Erglis A.,
- Doh J.-H.,
- et al.
- Nørgaard B.L.,
- Leipsic J.,
- Gaur S.,
- et al.
- Kim H.Y.,
- Lim H.-S.,
- Doh J.-H.,
- et al.
- Taylor C.A.,
- Fonte T.A.,
- Min J.K.